All right. Good morning, everyone. Thank you for joining us at the William Blair conference. My name is Margaret Kazer Andrew, and I am the analyst here that covers Ceribell. There is a complete list of research disclosures and conflicts of interest at williamblair.com. Before we hit off the presentation, I'll maybe start off with a 30-second pitch here. Ceribell was one of our feature stocks going into this week, and it's really around the fact that they are launching their own AI-driven product and software solution into the marketplace and in the early innings of that.
While I'm sure some of the audience have tried AI, like it, some of the audience maybe tried AI, didn't like it, are questioning it, the fact is that AI is going to make a tremendous amount of difference and change the landscape of healthcare, at least in our opinion, certainly over the next two to five years. Frankly, it's doing it today, and Ceribell is a wonderful example of that. With that, I'm going to turn it over to Jane and Scott, the CEO and CFO of Ceribell.
Thank you, Margaret. Good morning, everyone. So I'm the Co-founder and CEO. I would like to walk you through the Ceribell story. I first show you the overview of the company, then the clinical unmet need we're trying to address, our product solution, then our commercial performance as well as the business model, and quickly our future pipeline and this year's focus. With that, at the highest level, Ceribell invented a novel neural monitoring platform. It has both the hardware as well as the AI and algorithm. The hardware you'll see enables nursing, really anyone on the bedside can set up EEG or brainwave in a few minutes, in contrast to waiting for hours and days, which we'll walk you through. Also have, as Margaret mentioned, the AI-powered seizure detection algorithm, which I'll walk you through.
With that, the AI is only focused on seizure for now, but have future potentials. Our current focus is seizure detection in acute care, that's ICU and ED, and that in the U.S. alone translates to more than $2 billion TAM. We also have a very strong pipeline, mostly using EEG signal and AI for indications other than seizure, and that including delirium, stroke, and other indications that, of course, will significantly increase our TAM. Last quarter, we did $20.5 million. That translates to a 42% growth year over year. We are currently trending towards 600 hospitals, and our company-wide gross margin is 88%. I mentioned the current focus is seizure detection. Often when we think about seizure, we think about epilepsy. A few things come to mind. A, you can see the symptom. Patient has convulsion. B, these seizures are usually transient.
Usually, patients themselves will stop seizing after a few minutes. Last thing is this is a epilepsy is a devastating disease state, but it's usually not life-threatening. When you think about seizure in the ICU and ED, it's the complete opposite on all the three dimensions. First of all, there are very little symptoms. You don't see patients having convulsion. Many of these patients are intubated, comatose. You have to have an EEG to know patients are seizing. B, patients don't just stop seizing after a couple of minutes. Patients can seize from hours to days. Last one, therefore, it leads to very high mortality and morbidity. It's very common. A third of neuro patients in ICU have seizure. It also goes beyond neuro. If you look at the indication, it goes from patients have status epilepticus convulsively.
You're going to hear me saying status epilepticus. If it's the first time I'm hearing Ceribell talk, it's likely you never heard of it before. It's a medical emergency defined as seizure lasting for five minutes or longer. If a patient had a convulsive long seizure, 50%, they will have another non-convulsive seizure, post-TBI, traumatic brain injury, post-stroke, post-neurosurgery. It goes beyond neuro patient, sepsis patient, post-cardiac arrest patient. Somewhere between 10%-20%- 50% of these patients have seizure. When seizure happens, 92% of them have no symptoms. Therefore, you have to have an EEG to monitor these patients. We often say time is brain for stroke. Our doctors say this, time is brain for seizure in ICU. The time, and here's why. On the left side, you see how mortality goes up significantly just with hours of delay.
If a patient seizes for 10 hours-20 hours, mortality is about 33%. If a patient seizes more than 20 hours, mortality goes to 85%. You might be wondering which modern ICU would allow their patients to seize for 10 hours -20 hours. Pretty much any hospital you can walk to now if they do not have Ceribell, right? I will explain why. It is very common. The good news, similar to stroke, is the treatment is readily available and can be very effective if patients are treated early. Again, you see on the left side, on the right side, if you treat the patient within the first 30 minutes of seizure onset, 80% of patients respond. Usually, first-line medication is the best chance for patients to stop seizing, then case closed for this patient.
If you just delay that treatment by a couple of hours, patients stop responding. That line just keeps dropping. Early detection, therefore early EEG, is the most important factor for seizure management in the acute care setting. The guideline already agreed with this. More than 10 years ago, Neuro Critical Care Society recommended if you worry a patient has status epilepticus, long seizure, you need to EEG arrive on the bedside 15 minutes-60 minutes. Later, level one recommendation from cardiac arrest from AHA, American Heart Association, and post-stroke patient for seizure management from stroke societies. If you look at the first one with specific timeline, 15 minutes-60 minutes, before Ceribell, not a single hospital on the planet can be compliant to this guideline. Here's why. Before Ceribell, the current, in some ways, still the current gold standard, we are changing that.
The gold standard is the conventional EEG, which was invented literally 100 years ago. During the past 100 years, not much has changed. Of course, we digitalized it. It was designed for epilepsy diagnosis in the clinical settings where a patient can wait for weeks or months. It was not designed for acute care setting where every minute, every hour counts. There are three intrinsic bottlenecks. First of all, you have to have an EEG technician to set it up. The entire hospital, usually large seizure center, has 5-10 EEG tech. Most community hospitals have 1 or 0 EEG tech. They're on site Monday to Friday, 9:00 A.M. to 5:00 P.M. Often you wait for hours or even days to get them. When they get to the bedside, it takes them half an hour to set up.
Even when they set up, EEG recording does not help anyone. You need an epileptologist often or neurophysiologist to read it. Again, a large teaching center, you are talking about 5-10 physicians out of thousands can actually read EEG. They are almost never at the bedside. Community hospital, you are talking about one or two physicians. When you combine all these bottlenecks together, the reality, oh, sorry, the last one is nobody continues to monitor it. That means you stare at the screen nonstop. With the current neurology shortage, it is almost humanly impossible. The reality of getting a conventional EEG instead of 15 minutes-60 minutes, it always takes hours and often days for all the bottlenecks. What does this mean to the bedside? Let's say you are an ED or ICU physician, and you worry a patient has seizure.
They have a couple of options. The first option is you wait. You order EEG because you do not want to just treat the patient because you do not know whether or not the patient has seizure. If you wait, you remember the mortality goes up every hour. That is a very uneasy decision. Many doctors go with the second. They just empirically treat while they are waiting for the EEG. The problem is that the first-line treatment is a very large dose of benzo. It suppresses the patient airway. Usually, you also have to intubate the patient as you empirically treat. It is commonly happened. Multiple studies show 40%-50% of the time that is what doctors do. Patients end up having no seizure. The last one is barely an option. You transfer the patient out. Delayed suboptimal care. Very expensive for the hospital.
You lose the revenue for the hospital. If we zoom out, let's talk about what this means. Before Ceribell, if you are first time in Ceribell, you've probably never heard of status epilepticus. You've heard of stroke, sepsis, and cardiac arrest. Here's just a simple comparison. Look at the mortality rate. Very similar. Look at age impact. On average, status patient is 20 years younger, right? If you go to the hospital, all the administrators will tell you, we have too many protocols on sepsis, cardiac arrest, and stroke for right reasons. Almost no hospital has a proper protocol for status epilepticus because you can't afford to put a protocol that you're going to fail yourself if you don't have the tool. That's what we are changing. Instead of waiting for hours and days, Ceribell system allows setting up EEG in just a few minutes.
In a few minutes, again, the algorithm, the AI algorithm would analyze the EEG signal and immediately tell the bedside physician, is your patient seizing? Is your patient status epilepticus or your patient seizure-free? Now, the bedside would know immediately what to do. AI doesn't stop because seizure is dynamic. Patient can seize for five minutes, stop seizing, and seizure come back. It's almost like EKG. You need to continuously monitor the heart. The AI never gets tired. It continues to monitor every 10 seconds and alert bedside whenever a patient is in danger of seizure. I walk you through how we do that. On the left side is the hardware. The top part is a single patient use disposable headband.
That allows nurses, residents, anybody on the bedside with 10 minutes-20 minutes training, they can set up EEG in a few minutes. That is equivalent signal quality as conventional EEG. The recorder, of course, records the EEG, displays the EEG and other information, and streams the signal through the hospital Wi-Fi that is HIPAA compliant, FDA cleared, to our portal. The portal allows neurologists to log in to their phone or iPad or computer anytime, anywhere to review EEG in real time. Believe it or not, it is the first time many neurologists were able to read EEG on their phone. All these three components already replace the majority of the functionality of conventional EEG. The AI component is what we call Clarity.
That is the machine learning algorithm we started developing eight years ago, just two years after we founded the company when nobody was talking about AI because we believed that's the right tool. Let me walk through what Clarity does and what it is. On the left side, you see all these complicated lines on the top part. That's the EEG tracing. That's what neurologists see on our device or on conventional EEG as you can appreciate how complicated it is and why very few people can read it. The bottom part, without knowing what it is, that's the Clarity seizure burden. I'm going to explain to you later, but without knowing anything, where you want to click, where it's red, and that's the peak. That's where seizure happens. This would tell you during the entire recording, when does the event happen?
When the neurologist goes there, they know where to focus. It makes their reading much more efficient. On the bedside, when a patient is getting close to status epilepticus, seizing, the device will turn red and start beeping, also sees the trend. That is a lot of information. I'm going to share a case study with you so you can see how all this takes into action and how it compares to standard of care. This is a real patient case. The bottom part you see is the 1:00 A.M. Majority of the hospital would never get EEG at 1:00 A.M. to start with. This patient has some kind of twitching. It's not a confirmative seizure. The nurse worried. ICU doc ordered it. They set up EEG at 1:00 A.M. In most of the hospitals, even you can set up EEG at 1:00 A.M.
Very few hospitals have neurologists to read for you right away. In this case, within 5, 10 minutes, the device turned red and started beeping, saying your patient is likely to be in status epilepticus. It's continuous seizing. You see a little pink needle on Ceribell that. That's the real bedside annotation. The nursing treated the patient right away, of course, with the physician order. I mentioned you read the chart. Not every single patient responds to the first-line medication. Again, standard of care, you treat, you wait till the next day. Neurologist reading. In this case, we say, hey, the patient didn't respond. That's the second alert you see. They escalate the treatment. You can see the seizure burden starts dropping. Patient became seizure-free around 3:00 A.M. for about two hours. Here, continuous monitoring is important because seizure is dynamic.
This patient returned to full-blown status again at 4:00 A.M. Again, take them 10 minutes, treated the patient, seizure getting under control. I asked, which modern ICU would allow your patient to seize 10-20 hours? This patient, majority of hospital, she would have seized the entire night if she's in a good hospital. If she's in a common community hospital, Friday afternoon, she would have seized two-three days. If she gets transferred out, she could seize three-four days, right? This is what we change. If you think about mortality, morbidity, this person's life, patient's life is potentially completely transformed and impacted. That's the story. Here's the data. We have done multiple studies showing how often physicians change their treatment decisions based on Ceribell. 40%-50%. Multiple studies.
We didn't show all the data from community hospital ICU to sophisticated neuro ICU to the emergency department in teaching centers, non-teaching centers over and over again. 40%-50% of the time, once doctor used Ceribell, they decided, oh, I thought I'm going to treat. I'm not going to treat. I thought this patient didn't have seizure. This patient actually has seizure. That's a huge portion, right? Not surprisingly, that translates to shorter lengths of stay because if a patient has seizure, you want to treat early. If a patient didn't have seizure, you don't want to intubate the patient. We also show we reduce unnecessary patient transfer due to seizure by 94%-100%. This really allows patients to stay in their own hospital. The hospital got to keep the patient as well as the revenue. The strongest publication came out last year.
This is a SAFER trial. That's the 4.1 days length of stay reduction association. If you compare Ceribell cohort with conventional EEG, if you're interested to learn deeper, I highly recommend you read that paper. Clinical evidence, clinical impact is very strong. How about health economics for the hospital, especially these days? It's equally strong. We both help hospitals to appropriately bill for their patient as well as reduce cost. On the reimbursement front, we qualify for most of the existing EEG CPT codes, especially the professional fee that inpatient qualifies. The technical fee, many inpatient doesn't qualify in general, not related to Ceribell. We also identify often major complication comorbidity, which can significantly increase the DRG code because the patient is in a more complicated disease state due to seizure.
As I mentioned earlier, hospitals can reduce the patient transfer just for EEG so they will be able to keep their reimbursement and their revenue. We also received the breakthrough designation for our AI algorithm. Based on our knowledge, it's probably the first and only. It's now the very first few ICU monitoring devices actually received that breakthrough. In conjunction with that, we received the new technology add-on code. That's $913 for eligible CMS patients just for Ceribell usage. On the right side, I show that reduced length of stay, one day in the ICU costs hospital $4,000-$5,000, four days $20,000, and our device costs less than $1,000 overall. Also reduced hospital dependence on highly specialized EEG tech. There is a national shortage of them. Hospitals often struggle hiring them, as well as optimize the productivity and the work-life balance for the neurologist.
We have published 35-plus peer-reviewed journals as well as trending close to 100 abstracts. Let me switch gear and talk about commercialization. We have two commercial revenue streams. One is the AI component. That accounts for 25% of our revenue. That is the monthly subscription fee hospitals pay us. I know AI is a buzzword now, but we actually started our commercialization around 2019, including AI. That is, hospitals actually pay for it since day one we launched Clarity. This is, of course, very much recurring revenue. The other 75% is the disposable single patient headband. For every single patient, we charge the headband. That is about 75% of the total revenue. Our sales force has two arms. One is the territory manager. They are responsible for account acquisition, acquiring new hospitals. The second team is called the clinical account manager, the CAM.
The CAM are largely focusing on existing accounts, some portion of maintenance, but a big part of the effort is driving utilization in the existing accounts go deeper, which we can talk more. Jointly, for the first 90 days or even longer, they launch these accounts together. This ensures that the territory managers not only just focus on quantity, but also the quality of these accounts because they get compensated by the quality of these launches. It also ensures that the relationship and knowledge of these accounts get handed over to the CAM organization. As you can see, the majority of our sales force focuses on maintaining the revenue. The majority of their time are focused on growing the revenue. There are a few unique traits of our business model. The first one is it's highly sticky. We have a very low attrition rate.
Also, the usage and reordering is very consistent. When we launch an account, we do not just train a few physicians or a few technicians. We train dozens, sometimes hundreds of physicians and always hundreds of nurses from ICU, often to the emergency department to step down units. We really integrate it to how hospitals operate. If, as future, we have now, we have some competition, the switching cost is very high. The second part is, as I mentioned earlier, both our sales forces are focused on the growth. The unique about TAM, the territory acquisition team, once they close the account, they hand over. They can keep hunting, right? That is our most tenured sales so that they do not run into capacity issues. The last one is reoccurring revenue, as well as the high gross margin.
This chart kind of shows, especially the recurring revenue. As you can see, despite all the challenges, micro in the external environment, we do not have a single quarter that is lower than the previous quarter. You see very steady growth. It is compounded by both adding more accounts as well as growing more usage in the existing accounts. As I mentioned earlier, our overall TAM in seizure detection alone is more than $2 billion. How do we get there? It is a combination of the number of patients, that is the single headband use, as well as the number of accounts. As we went public in Q4 last year, at the time, we said there are about 5,500 hospitals in the U.S. that should be our TAM. That is all the short-term acute care, as well as critical access centers, as well as the freestanding.
During the past couple of quarters, we further expanded our TAM, one by receiving FedRAMP that qualifies us to operate cloud-based, therefore, our Clarity AI algorithm in all the VA facilities. As we shared in the last earnings call, we were very excited. We already launched the first cohort of VA successfully. That expanded our TAM by about 200 hospitals. The other expansion is very recent news that we have received Clarity expansion from 18 FedRAMP adult population to pediatric one year and older. Again, this is the first and only seizure detection algorithm ever received from such a young patient group. That would open the door for us for the Children's Hospital. That's about 280 hospitals. During the past couple of quarters, we already expanded our hospital TAM to close to 6,000.
On the patient population front, we looked at the ED and ICU patients, about 3 million of patients at reasonably high risk of seizure, and then they need to be monitored. If you apply this with RSP, you'll get about $2 billion. I'm not going to walk through line by line. Overall, we continue to focus on our core market. That's really focused on seizure detection in acute care settings. We continue to drive account acquisition. We continue to drive the utilization in our existing accounts. We continue to drive more clinical evidence to influence the guidelines. The upside is also for us to look into our pipeline and potentially OUS market, which I cover more. We think about our future that's now reflected in this slide in really three horizons.
The first horizon we are in now is seizure detection in acute care setting. That's about more than $2 billion. The second horizon is still focusing on acute care, but it goes beyond seizure. I'll give you an example. Let's say if you are a cardiac surgeon, you did a major cardiac surgery, and your patient didn't wake up in the window you expected. Now, what's your guess? Patient has a small but existing probability of having a stroke. Or patient is in non-convulsive seizures. Or patient is over-sedated during the surgery. The timeframe, treatment pathway, urgency is completely different. Even in the best center, you don't really have a way to differentiate that. EEG has been proven to be the biomarker, of course, for seizure, for delirium, for stroke, almost for the majority of the neurological abnormalities in the ICU and emergency department and beyond.
That's what we're aiming for for the midterm, is to have EEG to differentiate potentially stroke and seizure and other neurological indications in the ICU that make EEG a new vital sign. The third horizon, we can go potentially even beyond acute care setting. If you go to home monitoring or other outpatient clinics, if you search any neurological psychiatric disorder, ADHD, depression, OCD, and you put EEG as a biomarker, somewhere someone has done research showing that EEG is the biomarker. We are in the position that makes this possible because, first of all, before Ceribell, you can't even get EEG for seizure as hardware. Forget about the other indications. Most of these are academic research. We changed that. Second, to really use AI and EEG, you need a large database. You need to really make sure they're well labeled.
You need deep know-how of how these signals work because these are time-series signals. It is very different than large language models or image models, right? We have that deep know-how. With that, we are very excited about what we can do. Specifically for this year, on the seizure management front, as I mentioned, we will continue to grow our account acquisition as well as utilization in the existing account. At the beginning of the year, we said we also continue to grow in the VA system. Beginning of the year, we said we are going, we submitted pediatric Clarity in Q4 last year. We already gained FDA clearance on pediatric. We also plan to submit FDA clearance or application for neonate Clarity, as well as using EEG and AI for delirium. All that submission, we plan to happen this year.
We're very excited about the impact there. We are also continuously investing in existing platforms. For example, we potentially would start offering videos because in some of the situations, it's very beneficial for neurologists to see patients on video besides EEG. We'll also start offering EMR integration because that can significantly save further nursing time and help hospitals to bill and the patient record in a much frequent way, in a much easier way. We also are providing more product features for physician notification so physicians can decide when they want to get notified, at what time, on what patient population. All this will further improve user experience and increase the stickiness. With that, this is a platform and technology that actually saves lives every day as we are talking. It potentially significantly improves patients' clinical outcome as well.
It helps hospitals to save money, keep their patients in reduced length of stay. We have developed a proven approach for commercialization in account acquisition and really also start early success in driving utilization. The revenue is highly recurring with very high margin, significant TAM already U.S. alone, and further opportunity to expand our TAM, help more patients, and also drive the commercial success. We are very excited. Thank you for your attention. We are happy to take any questions
We have time for, oh, perfect. Please.
Yeah. So it's usually the subset of the patient population you see here. And then to estimate the TAM, we look at the guideline and work with the physicians. So one example, there is a level one AHA, American Heart Association guideline on post-cardiac arrest patients. If they are still comatose, they require, the society recommends rapid, prompt EEG monitoring for seizure management, right? So then we look at all the cardiac arrest patients, how many of them are comatose after the initial survival. And then we use that as a population, not the entire cardiac arrest patient as the TAM calculation.
Correct. Correct.
It's less about indication. So on the FDA indication front, doctors can order this for anyone. There is no patient population constraint. It's more from a clinical practice perspective. Doctors want to follow the guideline. Sorry, I cut you off. What is that?
Yeah.
Perfect. With that, we're out of time. Thank you for the question. The breakout will be in Burnham A upstairs. Thank you, guys.